Comparison of 2-year weight loss trends in behavioral treatments of obesity: diet, exercise, and combination interventions.

Overview

abstract

OBJECTIVE: The effects of three cognitive-behavioral weight control interventions for adults were compared: diet only, exercise only, and a combination of diet and exercise. This article reports 2-year follow-up data. DESIGN: The three interventions were compared in a randomized, experimental design. SUBJECTS: A total of 127 men and women who were at least 14 kg overweight (according to height-weight tables) were recruited from an urban community and assigned randomly to the experimental conditions. INTERVENTION: The dietary intervention was a low-energy eating plan adjusted to produce a 1 kg/week loss of weight. The exercise component involved training in walking and a home-based program of up to five exercise periods per week. There were 12 weekly instructional sessions, followed by 3 biweekly and 8 monthly meetings. All sessions were led by registered dietitians. OUTCOME MEASURES: Changes in body weight. STATISTICAL ANALYSES: Analysis of variance for weight changes and repeated measures analysis of variance for weight change trends. RESULTS: At 1 year, no significant differences were noted among the three groups. The diet-only group lost 6.8 kg, the exercise-only group lost 2.9 kg, and the combination group lost 8.9 kg (P=.09). During the second year, the diet-only group regained weight--reaching 0.9 kg above baseline; the combination group regained to 2.2 kg below baseline; and the exercise-only group regained slightly to 2.7 kg below baseline (P=.36). Repeated measures analysis of variance showed a group-by-time interaction (P=.001); data for the dieting groups best fit a U-shaped regain curve (P=.001). APPLICATIONS: The results suggest that dieting is associated with weight loss followed by regain after treatment ends, whereas exercise alone produced smaller weight losses but better maintenance. The large outcome variability and unequal difficulty of the regimens across groups limit the generalizability of the findings.